Line data    Source code

       1              : /**
       2              :  * Copyright (c) 2025 Huawei Technologies Co., Ltd.
       3              :  * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
       4              :  * CANN Open Software License Agreement Version 2.0 (the "License").
       5              :  * Please refer to the License for details. You may not use this file except in compliance with the License.
       6              :  * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
       7              :  * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
       8              :  * See LICENSE in the root of the software repository for the full text of the License.
       9              :  */
      10              : 
      11              : #ifndef HCCL_ORDER_LAUNCH_H
      12              : #define HCCL_ORDER_LAUNCH_H
      13              : 
      14              : #include <unordered_set>
      15              : #include <map>
      16              : #include <memory>
      17              : #include "hccl/hccl_types.h"
      18              : #include "stream_pub.h"
      19              : #include "hccl_common.h"
      20              : #include "local_notify.h"
      21              : #include "adapter_rts.h"
      22              : #include "log.h"
      23              : 
      24              : namespace hccl {
      25              : struct OrderLaunchResMgr
      26              : {
      27              :     u64 context;
      28              :     Stream opbaseStream; // 单算子模式使用
      29              :     Stream aclgraphStream; // aclgraph模式使用
      30              :     bool resValid;
      31              :     bool contextInitialized; // 标记context是否已初始化，一旦初始化就不会变回INVALID_U64）
      32              : 
      33            0 :     OrderLaunchResMgr() : context(INVALID_U64), opbaseStream(), aclgraphStream(), resValid(false), contextInitialized(false) {}
      34              : 
      35              :     // 检查context是否匹配，不匹配时需要更新context但保留stream
      36              :     bool IsCtxMatch(u64 currentContext) const {
      37              :         return context == currentContext && resValid;
      38              :     }
      39              : 
      40              :     // 更新context
      41            0 :     void UpdateContext(u64 newContext) {
      42            0 :         if (context != newContext) {
      43            0 :             HCCL_INFO("[OrderLaunchResMgr] context updated: [0x%llx] -> [0x%llx]", context, newContext);
      44            0 :             context = newContext;
      45            0 :             resValid = false; // context变化认为资源暂时无效
      46              :         }
      47            0 :     }
      48              : 
      49              :     // 设置context为已初始化状态（一旦调用，context就不会再变回INVALID_U64）
      50            0 :     void MarkContextInitialized(u64 newContext) {
      51            0 :         if (context == INVALID_U64 || context != newContext) {
      52            0 :             HCCL_INFO("[OrderLaunchResMgr] Mark context initialized: [0x%llx] -> [0x%llx]", context, newContext);
      53            0 :             context = newContext;
      54            0 :             contextInitialized = true; // 标记为已初始化
      55            0 :             resValid = false;
      56              :         }
      57            0 :     }
      58              : 
      59              :     // 析构资源
      60            0 :     void DestroyResources() {
      61              :         // 析构资源位置预留，后续根据需要实现
      62            0 :     }
      63              : };
      64              : 
      65              : class OrderLaunch {
      66              : public:
      67              :     static OrderLaunch& GetInstance(s32 deviceLogicID);
      68              :     HcclResult RegisterOrderLaunch(const std::string &group);
      69              :     HcclResult UnRegisterOrderLaunch(const std::string &group);
      70              : 
      71              :     HcclResult AclgraphLaunchInOrderToOrderStream(std::string &group, const Stream& kernelStream,
      72              :         std::shared_ptr<LocalNotify> notify0, std::shared_ptr<LocalNotify> notify1, u32 timeOut, HcclRtEvent event);
      73              :     HcclResult AclgraphLaunchInOrderToKernelStream(std::string &group, const Stream& kernelStream, HcclRtEvent event);
      74              : 
      75              :     HcclResult OpbaseLaunchInOrder(std::string &group, const Stream& kernelStream,
      76              :         std::shared_ptr<LocalNotify> notify0, std::shared_ptr<LocalNotify> notify1, u32 timeOut);
      77              : 
      78              :     HcclResult HcomLaunchInOrder(std::string &group, const Stream& kernelStream, u32 graphId,
      79              :         std::shared_ptr<LocalNotify> notify0, std::shared_ptr<LocalNotify> notify1, u32 timeOut);
      80              : 
      81              :     HcclResult SetHcomStream(u32 graphId, const Stream& hcomAttachedStream);
      82              : 
      83              : private:
      84              :     OrderLaunch();
      85              :     ~OrderLaunch();
      86              : 
      87              :     void DestroyRes();
      88              :     HcclResult LaunchInOrder(std::string &group, const Stream &kernelStream, const Stream &hostOrderStream,
      89              :         std::shared_ptr<LocalNotify> notify0, std::shared_ptr<LocalNotify> notify1, u32 timeOut);
      90              :     HcclResult InitGroupCtx(const std::string &group);
      91              :     HcclResult EnsureOrderStreamForGroup(std::string &group, u64 context, Stream &orderStream);
      92              :     HcclResult GetCurrentContext(u64 &currentContext);
      93              : 
      94              :     std::mutex streamMutex_;
      95              :     std::unordered_map<std::string, u64> groupCtxMap_; // group -> context 映射
      96              :     std::unordered_map<u64, std::unordered_set<std::string>> contextGroupsMap_; // context -> groups 映射
      97              :     std::unordered_map<u64, OrderLaunchResMgr> contextResMgrMap_; // context -> OrderLaunchResMgr (stream由context管理)
      98              :     std::unordered_map<u32, Stream> hcomStreamMap_; // 图模式使用
      99              :     bool initialized_ = false;
     100              : };
     101              : }
     102              : #endif